Liquid metal cooling

[Bal`thzar]

ok so a computer manufaturer is doing some testing with liquid metals for processor cooling. From what I can see they are injecting a metal with the same liquid properties into a closed heatpipe system. the liquid metal doesn't become a solid until it hits in the neighborhood of -50c. I am not sure how this will be cooled in the heatsink\heatpipe system I am working on getting that information, but I was thinking something like this would work well with a small phase change system where the phase change element woudn't have direct contact with the CPU. I am thinking that this non direct contact is safer for the CPU's

I am not a technical person when it comes to phase change or liquid and some of what I have is fact and some conjecture.

I am looking for some insight and dialogue about this subject and the posibility of using it to OC

[Pf.Farnsworth]

What you said doesn't really make sense, to me anyway. All metals can be in a liquid state. What you mean so say is liquid around room temperature.

Do you want to use this metal as a refrigerant? If so then not really. For that it needs to boil at a low temperature, if it freezes at only -50C so it probably boils well above 0C, which makes it pretty useless as a refrigerant. That and I don't think it would agree with the compressor oil very well.

Do you mean to use it as a thermal interface between the evap and the cpu? If so then what metal is it exactly, because arctic silver thermal interface paste everyone uses is silver by the most part which is one of the best heat conductors out there, hard to beat.

You might want to read up more about phase change based refrigeration cooling. A good understanding of the concept will answer most questions for you. And try to rephrase what you are trying to get a cross, I am a bit confused :)

[Shell]

I tried liquid gallium-tin alloys as a computer coolant once... if you actually bother making a system that can withstand the reactive metal and use a very powerful magnetic impellor pump, you can get many times the performance of what you'd normally see with any other form of coolant.

Gallium is 64x more heat conductive then water... and if you use a pure copper heatercore and silver waterblock with a gallium TIM you can see exceptional performance.

There is one problem though... if you want to use a chiller you'll have to be careful, some alloys can only take -20°C if they're under constant heat load while in motion. If the coolant were to turn into a solid, you'd destroy your pump and plug up your loop with solid metal... VERY BAD.

So make shure to have a couple flow indicators with a safety cut-out, otherwise you'll have to deal with exploding CPUs... and that'll put a hole in your mainboard.

And it eats Aluminum like mad, one drop and eat away a nice big chunk.

[greenmaji]

It sounds to me like the idea goes like this

CPU-> Heatpipe cooler with liquid metal inside the heatpipe -> DD that will hit -50C on load to cool the heatpipe on the hot side..

-50C load is a low temerature to achive with a single stage phase change unit, the best gasss would need to be in a vacume state to be able to pull this (based on my understanding of refigarants that are sutable for single stage phase change use)

In other words it would take a two stage cascade or two stage auto-cascade (both of wich is rather overkill but you could tune one or the other for a rather enorumus load at -50C without too much trouble)

HTH

btw.. I do not see how this will improve performance.. your adding more thernal interfaces to the system wich will be at the very least somewhat detrmental.. unless the change in phase inside the heatpipe improves load performance or creates more stable temperatures is the desired effect..

[Bal`thzar]

well its a Gallium, gold, Tin mix

I am waiting to see one of the heatsinks. Once i see one of them I can see whether they are using phase change or just using it to cool. I find this whole thing really interesting. Innitially I thought they were using it as thermal interface, but they arn't

[dominick32]

Rather than go through all of this time, money, and trouble for an absolute best of no better than -50*C? . Why not have a phase builder create a -100*C Cascade unit for $1400, plug it in, and play? Or better yet, for $500 buy a used modified Single Stage phase unit for -60*C idle, -47*C load ???

This liquid metal idea, holds absolutely no cost/benefit in my eyes.

[Bal`thzar]

Rather than go through all of this time, money, and trouble for an absolute best of no better than -50*C? . Why not have a phase builder create a -100*C Cascade unit for $1400, plug it in, and play? Or better yet, for $500 buy a used modified Single Stage phase unit for -60*C idle, -47*C load ???

This liquid metal idea, holds absolutely no cost/benefit in my eyes.


Dom I am just trying to discuss what this can do and some thoughts by people. If I wanted to build a traditional phase change I am sure I am capable of doing it myself, but this Liquid metal intrigues me so I am just doing some thinking out loud with people. I am not really looking for infomation on why it won't work or cost benefits. Just discussing basic information

[dominick32]

Dom I am just trying to discuss what this can do and some thoughts by people. If I wanted to build a traditional phase change I am sure I am capable of doing it myself, but this Liquid metal intrigues me so I am just doing some thinking out loud with people. I am not really looking for infomation on why it won't work or cost benefits. Just discussing basic information

I understand completely but my question to you is: If the liquid would turn to solid at -50*C what is the point of going any further with it? Unless you are really just discussing the methodology in creating something like this...

If I was to choose this or phase, it would obviously be phase. An evaporator will not stop working at -50*C, but this liquid metal would. Maybe I am dissecting your original post wrong, I am a little confused. Sorry if I mis-interpreted you.

Regards,

[greenmaji]

ehhh.. for change of phase it would have to change phases, the mix you were talking about whouldn't change phase @ any achivable preasure that I could possibly dream up..

Think about it.. what state would it be in when at room temperture? a liquid? a solid? how many delta C would it take to make it a vapor or a liquid at normal preasures?

What Im saying is it sounds like a better material for an experimental TIM then medium to go inside a heatpipe... They might "claim" that a TIM made of those metals are changeing phase from a solid to a liquid however.

[enz660]

ive been reading about one company thats doing testing with aliens. They hire illegal aliens to constantly refill their liquid nitrogen system. Good results so far... :P

[Shell]

Liquid metals are GREAT for TEC(pelt) cooling BTW.

You can your CPU at a nice frosty and the liquid could take up to... well... 200C before it turns into a very strange vapour... very strange...
A pretty awesome vapour if you have it contained in midair as a superchilled gas with a powerful electromagnet and use it as a form of "speaker".
My friend did that once... he never told me how exactly he did that. :confused:

Anyhoo! Liquid gallium is a great coolant for sticking a TEC waterblock onto anything. The TEC will run very efficiently with heat being dispersed soo rapidly and you won't have to worry about your coolant turning into a solid.

And besides constant motion or superchilling the liquid metal... the only other way I could see it being kept from turning into a solid was if you has an extremely strong magnetic field vibrating the metal at many Hz... but magnetizn does not work well with computers...

[LordFrank]

Shell, what do you use to cool Ga in a loop? Odd that a byproduct of aluminium would eat away aluminium.

There's not much worry about it freezing though, gallium has a strong tendency to supercool below its freezing point, so you'd proabaly have to do some seeding to cause it to become solid. I don't know all that much about it though...


EDIT:
Also you'd probably want it pure... which is kinda pricy, at a few bucks per gram.

[Pf.Farnsworth]

It sounds to me like the idea goes like this

CPU-> Heatpipe cooler with liquid metal inside the heatpipe -> DD that will hit -50C on load to cool the heatpipe on the hot side..


If this is what the OP means then no its not a good idea. What you are saying is lets cool the hot end of a heat pipe to -50C with a DD SS. Ok, then the logical assumption would be that you have a refrigerant in the heat pipe that takes very very low temperatures to condense, and hence even lower boiling point. So in essence you are making a cascade.

However first of all if the metal freezes at -50C then it probably boils well over 0C, which makes absolutely no sense to be used as a refrigerant for OUR purposes where we want to get colder. We already have simple and cost effective system that will go well below 0C, why would we want to stay warmer.

But even if this metal did boil at a very low temperature there is still no pint in a heat pipe as you could just make a normal cascade and use it in a second stage as a refrigerant.

Rather than go through all of this time, money, and trouble for an absolute best of no better than -50*C? . Why not have a phase builder create a -100*C Cascade unit for $1400, plug it in, and play? Or better yet, for $500 buy a used modified Single Stage phase unit for -60*C idle, -47*C load ???

This liquid metal idea, holds absolutely no cost/benefit in my eyes.

Dom I am just trying to discuss what this can do and some thoughts by people. If I wanted to build a traditional phase change I am sure I am capable of doing it myself, but this Liquid metal intrigues me so I am just doing some thinking out loud with people. I am not really looking for infomation on why it won't work or cost benefits. Just discussing basic information

-----------------------

Well thats Doms point, there is nothing really to discuss. I don't know why people without serious chemistry, physics and mathematical education insist on discovering the next great break through in so and so. Almost always given the person had proper education they wouldn't be posting what they are because they would know better. If it was this easy it would be done already. The only exception is when it has already been developed and is used somewhere just not for pc cooling.

As for the original post, there is absolutely no reason to do this because a simpler, cheaper, and more effective solution is available. End of discussion. I mean sure we can use a car as a mixer to make a cake and we can put a car on a tv instead of a jack stand, but there is absolutely no point as better in every way solutions are available for less cost.

[LordFrank]

It boils at 2204°C (and actaully that's give or take 200 degrees depending on who you talk to). It has one of the longest liquid ranges of any metal.

Melts at room temps, kinda like mercury, cesium, and rubidium.

Just fyi. heh


EDIT:
Conductivity | Electrical: 0.0678 10^6/cm ohm | Thermal: 40.6 W/mK



2nd EDIT, just for camparison:
Silver - 429 W/mK
Copper - 401 W/mK
Water - 0.561 W/mK
Liquid Nitrogen - 0.49 W/mK

(roughly)

[Pf.Farnsworth]

Exactly. If your cpu hits 2204°C you will have bigger problems to worry about lol